The intersection of telehealth and AI: How can they reinforce each other?
The intersection of telehealth and AI: How can they reinforce each other? Bill Siwicki
Telemedicine took a huge leap forward during the pandemic as more providers and patients embraced remote delivery of healthcare services.
Today, artificial intelligence-powered telemedicine tools that remotely diagnose patients and route them to the proper care setting have begun popping up; they are designed to increase access to healthcare services and assist providers in making informed decisions, ensuring timely and appropriate care.
As providers move toward increasing virtual care options across the care continuum, AI-powered patient triage and medical diagnosis can potentially create proper channels of care that not only bridge the gap in health disparities and promote access but also create a better patient experience and better health outcomes.
Piotr Orzechowski is CEO of Infermedica, an AI health company dedicated to improving preliminary symptom analysis and digital triage. We interviewed him to discuss the intersection of telemedicine and AI.
Q. What is the problem in telemedicine that you think Al is especially suited to address?
A. Artificial intelligence holds significant potential in enhancing telemedicine services. A primary area of application is aiding patients in identifying the correct level of care.
Our studies indicate approximately 75% of patients are uncertain about the appropriate care they require, leading to inefficient use of emergency rooms for minor issues or neglecting severe symptoms. AI's role is pivotal in guiding patients, advising them on when telemedicine is suitable, and suggesting alternative care options when necessary.
AI also significantly enhances telemedicine adoption. In a study conducted with one of our clients, we found 33% of patients initially planning in-person doctor visits opted for teleconsultation first, highlighting AI's role in guiding patient decisions. This helps in determining when telehealth is suitable and when physical examinations or further testing are necessary, thereby optimizing patient care efficiency.
Lastly, AI assists healthcare providers in managing extensive documentation tasks. In telemedicine settings, AI can interact with patients prior to consultation, gathering initial information and details about their primary concerns.
This is particularly relevant given the findings from the "I Cry, but No One Cares": Physician Burnout & Depression Report 2023, which revealed more than 50% of doctors experience burnout. AI's support in these areas can alleviate some of the workload, offering substantial relief to healthcare professionals.
Q. You suggest Al tools can remotely diagnose patients and route them quickly to the proper care setting. Please describe how this would work.
A. While our AI is not licensed for diagnosing, it plays a crucial role in influencing patient decisions. According to our data, as many as 77% of patients are likely to alter their initial intent after undergoing online triage.
AI tools help streamline the triage process, making it more user-friendly as well. It begins by collecting basic information like demographics and risk factors, followed by inquiries about the patient's primary symptoms, covering a wide age range from newborns to adults.
Using the initial data, the AI system prompts a series of questions that mimic the diagnostic approach of human clinicians, focusing on aspects like severity, duration, exacerbating factors and related symptoms. Typically, after 10 to 20 questions, the AI reaches a sufficient confidence level to offer a reliable triage suggestion.
The output is a results page outlining the recommended course of action, which may vary from self-care guidelines to advising a consultation with healthcare professionals or directing to urgent or emergency care. These recommendations also suggest the most suitable consultation mode, such as in-person or remote. Additionally, we provide educational articles to guide patients on interim care and home management.
It is crucial to note our AI tool does not diagnose nor dictate the patient's final decision. It serves as an educational aid, empowering patients to make informed choices about their healthcare. We firmly believe AI should support, not enforce, patient decision-making.
Q. Do you think Al tools are qualified today to do this kind of diagnosing?
A. Currently, AI tools are not authorized to diagnose patients. Despite the remarkable progress in generative AI, we must remain cautious about their practical application in healthcare. Our blood pressure cuffs are certified medical devices, and it's noteworthy that while AI tools possess significant capabilities, they are not subject to the same regulatory rigor.
It's critical to establish a robust regulatory framework to guide and set standards for AI-assisted diagnosis in the future. This includes addressing key challenges like ensuring maximum transparency in AI decision-making processes and tackling issues related to bias and inaccuracies.
I believe the ideal path forward is to position AI tools as optimal supporters for both patients and healthcare providers. This involves aiding patients in making informed self-care decisions and freeing healthcare professionals of repetitive tasks.
Q. What are the outcomes of incorporating Al into telemedicine in this manner? How would it, for example, help improve care and boost patient and provider satisfaction?
A. We collaborate globally with top telemedicine organizations, from health insurers to government health ministries. Our systems have triaged more than 15 million patients worldwide.
For instance, Solv's symptom checker, launched in October 2022, doubled in-network bookings and achieved an 80% satisfaction rating. Notably, 14% of users, initially not seeking care, booked appointments, reducing potential health risks and costs.
Diagnostikare, changing digital healthcare in Mexico, integrated our API for a symptom checker, enhancing patient care efficiency by 39%. This tool swiftly connects patients with doctors, shortening visits by 7.5 minutes and improving preparedness.
Our solution also led 39% of new patients to telemedicine for one payer, with 28% of those patients choosing teleconsultation over ER visits. The standardized triage process received high ratings from call center operators as well.
The HIMSS AI in Healthcare Forum is taking place on December 14-15, 2023, in San Diego, California. Learn more and register.
Follow Bill's HIT coverage on LinkedIn: Bill Siwicki
Email him: bsiwicki@himss.org
Healthcare IT News is a HIMSS Media publication.